Health Insurance Coverage Under the Affordable Care Act

Transcription

1 Estimates of the Impact of the Affordable Care Act on Texas Counties Prepared for Methodist Healthcare Ministries 2012 Final Report by Michael E. Cline, Ph.D. & Steve H. Murdock, Ph.D. Hobby Center for the Study of Texas, Rice University

2 Estimates of the Impact of the Affordable Care Act on Counties in Texas Prepared for Methodist Healthcare Ministries (MHM) Final Report by Michael E. Cline, Ph.D. Steve H. Murdock, Ph.D. Hobby Center for the Study of Texas Rice University Final Report April 2012

3 Executive Summary As of 2010, an estimated 23.7 percent of all Texans (5.9 million people) had no health insurance coverage, a larger percentage of population uninsured than any other state. The Affordable Care Act (ACA), the most comprehensive health care program passed by the Congress of the United States since 1965, is designed to expand health insurance coverage so that more people have access to health care. Despite its comprehensive nature there are likely to be persons who will not be covered by insurance even after full implementation of the ACA, either because they are exempt from specific provisions of the Act or because, for various reasons, they choose not to obtain health insurance. The differences in how many enroll in some form of health insurance will depend upon the efforts of the State of Texas and health care advocates to enroll people in public health insurance (Medicaid and CHIP) and how well the health benefits exchange is developed and marketed. Given the population in Texas in 2010, and had all of the provisions of the Act been implemented in 2010 and assuming a moderate scenario that increases enrollment in public and private health insurance, we estimate the following changes: A 3.0 million decline in the uninsured from 5.9 million to an estimated 2.9 million; 88% of Texans (and 87% of non-elderly Texans) would be enrolled in some form of health insurance (up from 77% and 74%, respectively); 165 counties would have larger percentage increases in the number of insured than the State as a whole (at a 15.3% increase); 34 counties would have larger proportions of the population remaining uninsured than the State as whole (at 11.6%) including: o The largest urban counties, including: Cameron (13.7%), Dallas (14.1%), El Paso (12.9%), Harris (13.9%), Hidalgo (15.0%), Tarrant (12.0%), and Webb (14.2%); o Many rural counties throughout Texas; o Counties in the South and West Texas border area. The percent and number uninsured decreases for all counties. Areas in Texas that will benefit most from the ACA are those counties where health insurance rates are already low. These counties are located primarily in rural areas of the state and in particular, areas in South and West Texas, and central city counties. After full implantation of the ACA, we estimate the following changes by county type: Average percent uninsured will decrease from 22.7% to 10.0% for rural counties, increasing the number of insured rural Texans by 395,000 people (16.8% increase) Average percent uninsured will decrease from 23.7% to 11.0% for central city counties, increasing the number of insured urban Texans by 2.0 million (15.8% increase); Average percent uninsured will decrease from 21.1% to 9.5% for suburban counties, increasing the number of insured suburban Texans by 587,000 (13.3% increase) These geographic differences in impacts are a result of differences in the socioeconomic and demographic characteristics of county populations and how the provisions of the Act will differentially impact specific demographic groups. Counties most impacted by the provisions of i

4 the Act will be those with larger proportions of their population in households with low to moderate income because these are the households who benefit most from two major provisions of the Act: an expansion of Medicaid coverage to persons in households with incomes less than or equal to 138 percent of the Federal Poverty Level (FPL) and, the development of a Healthcare Insurance Exchange with subsidized coverage for those persons in households with incomes less than or equal to 400 percent of FPL. Those counties with higher relative proportions of their population uninsured following full implementation of the ACA are those counties with larger numbers of immigrants, fewer governmental employees, and larger proportions of the population with low to moderate incomes. This report provides an overview of the estimated impacts of the ACA as passed and fully implemented and the methods used to estimate these impacts. ii

5 Table of Contents Introduction... 1 Overview of the Affordable Care Act... 2 Methodological Approach... 4 Population Impact Groups... 7 GROUP 1: ELDERLY (AGE 65 AND OLDER)... 8 GROUP 2: UNDOCUMENTED IMMIGRANTS... 8 GROUP 3: RECENTLY ARRIVED LEGAL PERMANENT RESIDENTS (LPR)... 9 GROUP 4: CHILDREN (AGE 0-18) IN HOUSEHOLDS WITH INCOMES LESS THAN OR EQUAL TO 200% OF FPL GROUP 5: CHILDREN (0-18) IN HOUSEHOLDS WITH INCOMES % OF FPL GROUP 6: CHILDREN (0-18) IN HOUSEHOLDS WITH INCOMES GREATER THAN 400% OF FPL GROUP 7: ADULTS (AGE 19-64) EMPLOYED IN GOVERNMENT GROUP 8: ADULTS (AGE 19-64) IN HOUSEHOLDS WITH INCOMES AT OR BELOW 138% OF FPL GROUP 9: ADULTS (AGE 19-64) IN HOUSEHOLDS WITH INCOMES % OF FPL GROUP 10: ADULTS (AGE 19-64) IN HOUSEHOLDS WITH INCOMES BETWEEN % OF FPL GROUP 11: ADULTS (AGE 19-64) IN HOUSEHOLDS WITH INCOMES GREATER THAN 400% OF FPL State Level Estimates of Population Groups and Estimates of Health Insurance Coverage Summary of Impact Estimates in Counties in Texas Works Cited Appendix A: Appendix B: iii

6 Tables Table 1: States with the Largest Share of the Civilian Non-Institutionalized Population Uninsured, Table 2: Health Insurance Coverage Pre- and Post-ACA... 4 Table 3: Estimates of Health Insurance Coverage for the Nonelderly Population Under ACA Table 4: Health Insurance Coverage Assuming Differential Impacts of the Affordable Care Act on Selected Groups and Three Different Policy Environments Table 5: Average Percent Uninsured for Counties by County Type Assuming Differential Impacts of the ACA on Selected Groups and Three Different Policy Scenarios Table 6: Change in the Insured and Uninsured by Alternative Scenarios and County Type iv

7 Figures Figure 1: Regions Used in the Estimation of Population Groups... 6 Figure 2: Estimates of Current Rates of Uninsurance for Counties in Texas Assuming Group Figure 3: Percent Increase in the Insured for Counties in Texas Under the Limited Policy Scenario (Compared to the State Increase in the Insured of 7.2 Percent) Figure 4: Percent Increase in the Insured for Counties in Texas Under the Enhanced Policy Scenario (Compared to the State Increase in the Insured of 22.8 Percent) Figure 5: Estimates of Rates of Uninsurance for Counties in Texas Under the Limited Policy Scenario (Compared to State Rate of Uninsurance of 18 Percent) Figure 6: Estimates of Rates of Uninsurance for Counties in Texas Under the Enhanced Policy Scenario (Compared to State Rate of Uninsurance of 6 Percent) v

8 Introduction In 2010, 5.9 million Texans had no health insurance coverage (U.S. Census Bureau 2010). This was the second largest number of uninsured among the 50 states (behind California) and the largest percentage of the overall population uninsured [see Table 1 (at 23.7 percent)]. The Affordable Care Act (ACA) 1, the most comprehensive health care program passed by the Congress of the United States since 1965 (when the passage of the Social Security Act established Medicare and Medicaid) is designed to decrease the numbers of the uninsured by creating avenues for obtaining affordable health insurance. The ACA accomplishes this by expanding public health insurance coverage (through expansions in Medicaid); establishing a market based health exchange market; offering individual subsidies; offering business tax credits, and adding tax penalties. Increasing health insurance coverage will enable more people to access health services thus increasing demand on the healthcare system. At the same time, despite its comprehensive nature, there are likely to be persons who will not be covered by insurance even after full implementation of the ACA, either because they are exempt from specific provisions of the Act or because, for various reasons, they choose not to obtain health insurance. The purpose of this research is to estimate changes in the number uninsured (and insured) for the State of Texas and counties within Texas. Table 1: States with the Largest Share of the Civilian Non-Institutionalized Population Uninsured, 2010 State Number (Millions) Percent Texas Nevada Florida Alaska Georgia New Mexico Oklahoma California Mississippi Louisiana Source: American Community Survey, 2010 During legislative deliberations and following passage of the ACA, several groups analyzed potential impacts of the ACA for the United States as a whole and for individual states (Buettgens and Hall 2011; Auerbach et al. 2011; U.S. Congress 2011). These studies predict that the percent of the non-elderly population insured in Texas would increase to from 87 percent (Buettgens and Hall 2011) to 93 percent (Auerbach et al. 2011). Using primarily economic based models, these studies incorporated assumptions about how actors (businesses, 1 Commonly referred to in general as the Affordable Care Act (ACA), the healthcare reform legislation as passed encompasses two official Acts of Congress passed in March 2010: the Patient Protection and Affordable Care Act (HR 3590) and the Health Care and Education Reconciliation Act of 2010 (HR 4872). 1

9 individuals, and governments) would respond to changes in health insurance costs with changes in health care policy (through expansions in public health care, subsidies to individuals and businesses, and tax penalties). These economic models are based on economic theory and empirical research regarding changes in state and federal health insurance policies (including post analyses of the expansion of public health insurance coverage through the Children s Health Insurance Program or CHIP and other state initiatives to increase health insurance coverage such as the Massachusetts health care reform law). These models are limited because their assumptions are derived from past actions by many actors (businesses, individuals, and governments) within the context of existing policy using assumptions about how those actors will change their behaviors under the new policy. In addition, some of these models are based upon empirical evidence derived from changes in policies in areas where health insurance coverage was already higher relative to other states. Given the complex nature of the provisions of the ACA, no model will be able to predict with absolute certainty the changes in the number and percent of the uninsured following full implementation of the related regulations and programs. In fact, few studies have undertaken an effort to understand the impact of the ACA for small areas within a state as is done in this analysis. In order to provide results that can be understood by a wider audience, we use population based or synthetic methods to estimate population groups categorized by their likelihood of enrolling in health insurance following full implementation of the ACA. Then we apply a range of assumptions about how each of these groups will change their insurance status as a result of the provisions of the Act. The uninsured (and insured) are then aggregated at the county and state level to produce estimates of the uninsured (and insured) according to three sets of scenarios: one in which limited efforts are made to enroll individuals in Medicaid or to establish a fully functioning Health Insurance Exchange; one in which modest efforts are made to encourage public and private health insurance enrollment; and finally, one assuming that most individuals will enroll in health insurance as a result of enhanced institutional efforts to encourage enrollment and individual mandates and incentives encouraging enrollment. At the statewide level, our analysis shows similar results to those previously prepared (Auerbach et al. 2011; Buettgens and Hall 2011; U.S. Congress 2011; Texas Health & Human Services Commission 2010) a change in coverage to 88 percent of the population being covered by some form of health insurance assuming modest efforts by the State and health care advocates to encourage public health insurance enrollment and increase access to affordable health insurance through a market based exchange. This report provides an overview of the major provisions of the Affordable Care Act (ACA); the methods used to estimate the number and percent of the population remaining uninsured assuming full implementation of the Act as passed; and an analysis of the statewide and county level estimates of, and changes in, the number of uninsured (and insured). Overview of the Affordable Care Act The Affordable Care Act (ACA) increases the number of people enrolled in public and private health insurance through a variety of incentives and penalties. Before providing an overview of the methods for projecting potential impacts, it is important to understand the provisions of the Act that will have the most significant impact to the largest number of people. This section provides such a summary but does not discuss all of the ways in which individuals will be covered following full implementation of the ACA. For a more comprehensive review of the provisions of the ACA, please see the companion report, Impact of the Patient Protection 2

10 and Affordable Care Act on Various Population Groups in Texas (Warren & Jahnke, 2010). The major provisions of the Act include the following: Health Benefits Exchange. People without access to employer-sponsored insurance (ESI) can purchase insurance through a market based insurance exchange. Premiums are subsidized through tax credits for households with incomes at or below 400 percent of the Federal Poverty Level (FPL). Insurance Mandates and Penalties. U.S. citizens and legal permanent residents must have qualified health insurance coverage or pay an income varying tax-penalty (with some exemptions). Medicaid Expansion. Individuals and families with incomes less than 138 percent of FPL will be eligible for Medicaid. Young Adult Coverage. Young adults age may stay on their parent s health insurance plan until their 26 th birthday. Tax credits for small businesses and nonprofits. Small businesses and nonprofits employing 25 or fewer full-time equivalent employees and having average annual wages under $50,000 can receive tax credits for covering at least half of the premium costs. Of the various provisions of the Act, two that will have the most profound impacts on changes in the numbers of individuals uninsured are 1) the expansion of Medicaid to nondisabled adults with family income at or below 138 percent of the FPL 2 ; and 2) changes in coverage as a result of the development of a health benefits exchange, with subsidized coverage for individuals in families with incomes at or below 400 percent of FPL (Cook, Dubay, and Garrett 2009; Holhan and Headen 2010; Warnke and Jahnke 2010). The first and second panel of Table 1 shows how children and adults are covered by health insurance prior to the full implementation of the ACA (i.e. the provisions of the Act effective on or before 2014). The third and fourth panel of Table 1 show how these same groups are covered following full implementation of ACA. Since the major provisions of the Act are based upon the income of an individual s family (or of an individual if single), the table shows how individuals falling within each income range will be covered. Healthcare provisions are established according to family income relative to the official Federal Poverty guidelines. These Federal Poverty Guidelines (also referred to as the Federal Poverty Level or FPL) are issued each year by the United States Department of Health and Human Services. 3 For illustrative purposes, the income equivalents for a family of four in 2011 are shown in this table. 2 Although the provision in the Act establishes a threshold of 133 percent, the effective standard eligibility will be 138 percent. Eligibility is based upon modified adjusted gross income of 133 percent with no asset or resource test plus a special adjustment of 5 percent. For more information about this, see: fpl-calculations-in-affordable-care-act. 3 See 3

11 Table 2: Health Insurance Coverage Pre- and Post-ACA Health Insurance Coverage Options for the Non-Elderly Non-Disabled by Income Federal Income Pre-ACA ACA Poverty Level Equivalent 1 Children Adults Children Adults Medicaid 2 Medicaid3 (Parents <12% FPL) Medicaid 4 Medicaid 138% FPL $30,843 Subsidized Exchange Subsidized Exchange CHIP Employer Sponsored Employer Sponsored Employer Sponsored Employer Sponsored Individual Individual Individual Individual CHIP 5 CHIP % FPL $30,843-$44,700 Employer Sponsored Employer Sponsored Employer Sponsored Employer Sponsored Subsidized Exchange Subsidized Exchange Individual Individual Individual Individual Employer Sponsored Employer Sponsored Employer Sponsored Employer Sponsored % FPL $44,701-$89,400 Subsidized Exchange Subsidized Exchange Individual Individual Individual Individual Employer Sponsored Employer Sponsored Employer Sponsored Employer Sponsored > 400% FPL >$89,400 Exchange Exchange Individual Individual Individual Individual 1 Family Income Equivalents for a Family of Four in Children Age 0-1 are eligible for Medicaid up to and including 185% FPL, Children Age 1-5 are eligible for Medicaid up to and including 133% FPL, Children 6-18 are eligible for Medicaid up to and including 100% FPL. 3 With the exception of pregnant women in families with income below 185% FPL, only adults with children in families below 12% FPL are eligible for Medicaid 4 Children Age 0-1 are eligible for Medicaid up to and including 185% FPL. All other children in families with incomes up to and including 138% FPL. 5 Asset tests for children in families with income % FPL. Methodological Approach In order to estimate the uninsured population following full implementation of the Affordable Care Act (ACA) for the State of Texas and counties within Texas, the population was classified into subgroups. These subgroups were defined according to the group s likelihood of being impacted significantly by the bill (either by gaining coverage due to specific provisions of the bill) or experiencing little or no impact as a result of the Act s implementation (because the group s rate of health insurance coverage is already high or low and the provisions of the Act are not likely to change the status of group members significantly). We did not estimate two subpopulations that are specifically identified by provisions of the Act (young adults who can be included under their parent s insurance policies and individuals in the high risk pool). Although these provisions will have an important impact on such individuals, these populations are difficult to estimate separately and in the aggregate their impacts will be minimal as compared to the overall impacts of the bill. As estimated here individuals in the categories used in this analysis are included within the defined population subgroups. After estimating population subgroups at the county level, we apply a state level estimate of current health insurance coverage for each group as well as three different assumptions in the level of institutional efforts to increase health insurance coverage and individual responses to those efforts. We use a population-based ratio (or synthetic) method to develop county level estimates of levels of insuredness in population subgroups. These methods are commonly used to estimate or 4

12 project socioeconomic characteristics of the population (Murdock et. al. 2010; Siegel 2002; Manton, Singer and Suzman 1993). The Integrated Public Use Microdata Sample (IPUMS) of the American Community Survey [(ACS) Ruggles et al. 2009] was selected as the primary data source for the estimation of the various groups. The ACS is a U.S. Census Bureau continuously conducted survey of the population of the United States that collects data on the socioeconomic characteristics of the population and of households. Approximately 1 in 11 households respond to the survey and the survey responses are weighted to population estimates developed by the U.S. Census Bureau. This is much larger than other sample surveys [such as the Current Population Survey (CPS)], which allow for greater accuracy in estimating population subgroups within small geographic areas within a state. The IPUMS is a modified version of the U.S. Census Bureau s Public Use Microdata Sample (PUMS), which is a one percent sample of individual records lacking any specific information that can be used to identify the individuals who responded. These records are identified only by their location within census designated geographic areas that consist of estimated total populations of 100,000 or more (areas called Public Use Microdata Areas or PUMAs). The IPUMS assign uniform variable codes across years for multiple data sets (which allows for pooling of data and estimating certain socioeconomic characteristics). In order to reduce sampling errors and develop reasonable estimates of the impacted population subgroups, we selected data from the 2005 to 2009 ACS. Like any sample survey, the ACS is subject to sampling error which can be compounded when trying to estimate small groups for small geographic areas. However, no other data set provides the geographic breadth of the ACS. The primary groups estimated included: 1) Undocumented immigrants; 2) Recently immigrated legal permanent residents (those who arrived in the United States in 1996 or after); 3) Adults employed in government; and, 4) Adults and children living within various levels above and below the official Federal Poverty level. These subgroups were estimated for Texas and for counties within the State. These methods were applied as indicated below. First, individual records within the IPUMS were classified by broad age groups (children- Ages 0-18, adults-ages 19-64, and the elderly-ages 65+), three race/ethnic categories (Hispanic, Non-Hispanic White, and All Others), and four income categories ( 138 percent of the Federal Poverty Level (or FPL), percent of FPL, percent of FPL, and > 400 percent of FPL). The adult, nonelderly population was further classified according to a combination of occupational/employment characteristics and immigration status. Individual records were classified as likely to be citizens, undocumented immigrants, or legal permanent residents based upon socioeconomic, demographic, and immigrant characteristics. In addition, legal permanent residents were also classified according to the year in which they immigrated to the United States. We first identified two specific immigrant groups: those likely to be undocumented and those who arrived in the United States in 1996 or later. Using class of worker, occupation, and industry variables, the remaining records not classified into these two immigrant groups were classified according to their occupational/employment characteristics: adults employed in government and all other adults. For the State and separately for each combined region, ratios of socioeconomic 5

13 categories to age and race specific groups were established. The region based ratios were then applied to counties within each region. In order to increase sample size, Public Use Microdata Areas (PUMAs) were combined. For most metropolitan central city counties, multiple PUMAs were combined to establish rates for the central city county. For the largest metropolitan areas (Dallas-Ft. Worth-Arlington and Houston-Sugarland-Baytown) the PUMAs surrounding the central county were combined in order to establish suburban estimates within the MSA. For the remaining regions, the geographic areas of analysis consisted of combinations of counties outside of these central counties and major suburban county areas. In order to increase sample size and decrease sampling error while still maintaining some variability in the resulting rates, these mostly non-metropolitan county regions were further divided into three major regions: West, East, and South Texas regions. These regions are created to approximate identifiable cultural regions within Texas where PUMA areas could be reasonably combined (Cline 2008). The regions are shown in Figure 1. Regional ratios for these estimation groups were applied to county age/race/ethnic specific population counts for counties in the respective regions. The resulting county estimates of these groups were then controlled to the state estimates. The policy impact scenarios that assume differential rates by population group were then applied to estimates of each group within each geographic area type (state or county). The use of a standard rate (i.e. the insurance coverage rate for each category for the state as a whole) allows us to understand how differential socioeconomic characteristics of the population within counties will impact the number of insured and uninsured. Figure 1: Regions Used in the Estimation of Population Groups Legend Regions Urban Counties West Texas South Texas East Texas Suburban DFW Suburban Houston 6

14 Population Impact Groups and Theirs Use in Estimation of Coverage As noted above in order to estimate the uninsured population in Texas following the implementation of the Affordable Care Act (ACA), the population was divided into groups defined on the basis of their eligibility for public health insurance (e.g. Medicaid, Medicare, CHIP, etc.) or subsidies and tax credits as a result of the Act and their likelihood of being covered by health insurance after the ACA is implemented. These population groups are identified and defined below. For each group we present summaries of the pre-legislation eligibility for public and private health insurance coverage including the current estimated health insurance coverage rates. The current health insurance coverage rates for each group (except the undocumented) are derived from a pooled sample of the 2008/2009 American Community Survey Integrated Public Use Microdata Sample for the State of Texas. The ACS began collecting information about health insurance coverage in 2008 and asked respondents to indicate whether or not individuals in their household were currently covered by health insurance. 4 This is different than the widely cited Current Population Survey which asks whether or not individuals have had health insurance at any time during the previous 12 months. The point-in-time estimates (as provided by the ACS) are a better measure of rates of insurance than the measure provided by the CPS (Davern, et al. 2009). Our overall approach for estimating the uninsured was as follows. First the total population was combined into eleven subpopulations delineated on the basis of the provisions of the bill. We then indicate how each group will be impacted by the ACA. In some cases, legislation will have limited impact on a given group. In others, provisions of the ACA will have a direct impact as a result of the legislation that expands public coverage or provides incentives or penalties that encourage people in the group to obtain health insurance. Other groups will be indirectly impacted as a result of expansions by related groups (such as children gaining coverage as a result of parents obtaining health insurance coverage). For each group, we provide the assumptions about health insurance coverage assuming full implementation of the ACA (i.e. the policies that will be in place by 2014). Any predictions about how the government, individuals, and businesses will react to the provisions of the bill are inherently difficult. In accordance with standard projection principles, we provide three alternative scenarios that assume differing rates of health insurance for each group. In the enhanced policy scenario, we assume (for most groups) that 98 percent of the population within a group is enrolled in some form of health insurance the same rate as that of persons age 65 and older (primarily a result of their coverage by Medicare). This scenario assumes an aggressive program of full implementation of the Medicaid expansion provisions of the Act, a fully developed health benefits exchange program, and assumes that the incentives, penalties, and mandates of the Act will encourage most to enroll. In the limited policy scenario, we assume that the Act will encourage many who were not previously enrolled in health insurance to enroll in health insurance, but that not everyone will enroll. Unless otherwise specified, this scenario assumes a change in the rate of insurance that is one fourth of the way from the current rate to near universal coverage (98 percent for most cases) for that specific group. While arbitrary, these changes in rates assume that some expansion in coverage will occur as a result of the elements of the ACA that encourage individuals to obtain coverage. These changes in rates are consistent with those that were found for children s health insurance coverage after the implementation of the 4 The U.S. Census Bureau made changes in their logical editing procedures for health insurance coverage types (methods to ensure consistency within a given record). The IPUMS variables on health insurance coverage are harmonized to be consistent across the 2008 and 2009 data years (for more information see: Lynch, Boudreaux, and Davern 2010; and 7

15 Children s Health Insurance Program (CHIP) the last major expansion in health insurance programs (Dubay et al. 2007; Dubay and Kenney 2009). The moderate policy scenario assumes a mid-point between the limited policy and enhanced policy scenarios. Each of the estimation groups is described below. Estimation Groups GROUP 1: ELDERLY (AGE 65 AND OLDER) Pre-Legislation Eligibility and Coverage: Primarily covered by Medicare. Post-Legislation Eligibility and Coverage: No change. Assumptions and Methodology for Estimation: Prior to enactment of the Affordable Care Act (ACA), health insurance coverage for the elderly population was nearly universal primarily as a result of Medicare. Estimates from the American Community Survey 5 indicate that approximately 98.2 percent of this population was enrolled in some form of health insurance (U.S. Census 2010). There are no changes in legislation that impact this population directly, thus all post-legislation scenarios will assume that 98 percent of the elderly will be enrolled in some form of health insurance. GROUP 2: UNDOCUMENTED IMMIGRANTS Pre-Legislation Eligibility and Coverage: Undocumented immigrants are not eligible for Medicaid or CHIP except during emergency situations. Although some undocumented immigrants may have employer-sponsored insurance (ESI) or purchase health insurance directly, most are not enrolled in health insurance. Post-Legislation Eligibility and Coverage: The ACA does not change eligibility for public health insurance for undocumented immigrants. Currently, undocumented immigrants may be enrolled in private insurance. After full implementation of the ACA, in addition to being excluded from public health insurance programs, these immigrants are excluded from incentives (in the form of tax credits and subsidies) and mandates (in the form of tax penalties). Thus, the legislation will have little to no impact on this population group. Assumptions and Methodology for Estimation: Logical edits were used to impute the legal status of the foreign born non-citizen population in Texas from the American Community Survey Integrated Public Use Microdata Sample (Ruggles et al. 2010). The algorithm used was similar to 5 The current health insurance coverage rates for each group (except the undocumented) are derived from a pooled sample of the 2008/2009 ACS IPUMS sample for the State of Texas. The ACS began collecting information about health insurance coverage in The ACS measures health insurance coverage at a point in time, unlike the CPS, which estimates health insurance coverage for any time during the previous 12 months. 8

16 that of Passel et al. and others who have estimated undocumented populations (Passel & Cohn, 2009; Passel & Cohn, 2008; Pew Hispanic Center, 2006; Passel & Clark, 1998; Passel, Van Hook, & Bean, 2004). Immigrants were classified as legal on the basis of other characteristics (such as year of entry, country of origin, occupation, etc.). Using this method, we estimated 1.5 million undocumented immigrants in Texas. This is within the 1.4 to 1.7 million range estimated by other analysts [see Table 2 (Hoefner, Rytina, & Baker, 2010; Warren, 2008; Passel & Cohn, 2009)]. The undocumented were distributed to counties by assuming that each county s share of the state s undocumented is the same as its share of the foreign born, non-citizen population in Texas who arrived in the United States in 1980 or later. Although it is not possible to determine the accuracy of such estimates because of a lack of an actual count of such persons, we believe that these estimates of the undocumented at the county level are reasonable and in line with our understanding of immigrant settlement patterns. The Pew Hispanic Center estimates that nationally, 41 percent of undocumented immigrants and 40 percent of Hispanic undocumented immigrants were insured in 2007 (Livingston, 2009; Passel & Cohn, 2009). Since the ACA does not impact this group directly, we assume that there are no changes in insurance coverage for this group as a whole. Thus, we assume that only 40 percent of this group will be covered by some form of health insurance after full implementation of the ACA and in all scenarios. GROUP 3: RECENTLY ARRIVED LEGAL PERMANENT RESIDENTS (LPR) Pre-Legislation Eligibility and Coverage: In Texas, legally authorized immigrant adults arriving after August 22, 1996 are not eligible for Medicaid or CHIP. Immigrants may purchase or obtain private health insurance (ESI or non-group coverage). Post-Legislation Eligibility and Coverage: No Change to public health insurance rules. All legally authorized immigrants will be eligible to purchase insurance through the exchange. Assumptions and Methodology for Estimation: Estimates of the immigrant population must distinguish legal permanent residents according to immigration year (through August 21, 1996 or after) and separately from the undocumented. The methods for estimating these different groups were described briefly in the previous discussion about the undocumented. Immigrants are asked about year of arrival on the American Community Survey, but are not asked about specific dates. Thus, we were limited in the estimates that could be developed for this population. We assume that all immigrants who were estimated to be legal permanent residents (based on the imputation methods that were used to distinguish the undocumented from other immigrant groups) and arrived in 1996 or later represent this group. All other legal permanent residents who arrived in 1995 or before were treated in the same manner as the remaining population groups. According to estimates derived from the American Community Survey IPUMS data for 2008/2009, an estimated 68 percent of this population was covered by some form of health insurance. While public coverage will not be available to this population, individuals in this group may purchase health insurance through the exchange. In the limited policy scenario, we assume no change in health insurance coverage for this group. In the enhanced coverage scenario, we assume that the coverage rate for this group will increase by 8 percent (to 76 percent due to expansions as a result of employer coverage and purchases through the exchange). 9

17 GROUP 4: CHILDREN (AGE 0-18) IN HOUSEHOLDS WITH INCOMES LESS THAN OR EQUAL TO 200% OF FPL Pre-Legislation Eligibility and Coverage: Children ages 1-5 and in families with income at or below 133 percent of FPL are eligible for coverage through Medicaid. Children ages 6-18 and in families with income at or below 100 percent of FPL are also eligible for Medicaid. All children in families with income at or below 200 percent of FPL are eligible for CHIP. Thus all children in families with income at or below 200 percent of FPL are eligible for some form of public health insurance. They may also be covered by individual purchased insurance or employer-sponsored insurance through their parents employer. Post-Legislation Eligibility and Coverage: All children in families with income at or below 200 percent of FPL remain eligible for some form of public health insurance. All children in families with income at or below 138 percent of FPL are eligible for Medicaid (regardless of age). Children at or below 200 percent of FPL are eligible for CHIP (with asset tests for children in families with income 150 to 200 percent of FPL). Children in this income category may also be covered under a parent s employer-sponsored insurance (ESI) or insurance purchased through the health benefits exchange. Assumptions and Methodology for Estimation: Although there may be shifts in the type of public coverage obtained (from CHIP to Medicaid) as eligibility criteria change for certain ages and income categories, there is no change in the overall eligibility for public coverage for this group. According to the American Community Survey, 76 percent of this population was covered by health insurance in 2008/2009. Some expansion in coverage is expected to occur due to the welcome mat effect whereby the currently eligible will enroll as a result of greater awareness of public programs as the ACA is implemented and as people enroll due to the individual mandate provisions. In addition, children currently eligible but not enrolled may be added as a result of expansions in public coverage for adults living in households with incomes at or below 138 percent FPL (for examples of this effect see Dubay and Kenney 2003). In the limited policy scenario, we assume a 6 percent increase in coverage for this group (to 82 percent insured). In the enhanced coverage scenario, we assume near universal coverage for this group (98 percent). GROUP 5: CHILDREN (0-18) IN HOUSEHOLDS WITH INCOMES % OF FPL Pre-Legislation Eligibility and Coverage: Children in households with incomes between 201 percent and 400 percent of FPL are not eligible for public health insurance. The primary source of health insurance for this group is through employer-sponsored insurance or through individual policies. Post-Legislation Eligibility and Coverage: No Change to public health insurance rules. Children in this income category may also be covered under a parent s employer-sponsored insurance (ESI) or through subsidized insurance purchased through the health benefits exchange. Assumptions and Methodology for Estimation: This group is not directly impacted by the legislation 10

18 since no expansions of public coverage occur. However, many children in this group who were not previously enrolled are expected to become enrolled in health insurance as a result of family access to health insurance through subsidized health insurance coverage in the exchange. In the limited policy coverage scenario, we assume a 4 percent increase from the level identified in the ACS (a rate of 88 percent for this group) in the limited coverage scenario. In the enhanced policy scenario, we assume 98 percent coverage. GROUP 6: CHILDREN (0-18) IN HOUSEHOLDS WITH INCOMES GREATER THAN 400% OF FPL Pre-Legislation Eligibility and Coverage: Children in households with incomes greater than 400 percent of FPL are not eligible for public health insurance. The primary source of health insurance for this group is through employer-sponsored insurance or through individual policies. Post-Legislation Eligibility and Coverage: No Change to public health insurance rules. In addition, these children live in households where subsidized exchange based policies will not be available. Assumptions and Methodology for Estimation: This group already has a high rate of health insurance coverage (95 percent). Because there are no changes in policy that would directly change the health insurance coverage for this group, we assume there are no changes in the rate of coverage for the limited policy coverage scenario. In the enhanced coverage scenario we assume 98 percent health coverage for this group. The remaining populations (adults age 19-64) are classified by household income category and employment status. Under the ACA, individuals are eligible to purchase insurance through the health benefits exchange and are eligible for certain subsidies based upon their household income. Persons in households with incomes less than or equal to 400 percent of FPL are eligible for subsidized insurance purchased through the health benefits exchange. In addition, penalties for not purchasing insurance vary according to household income. Current health insurance rates for these groups were estimated from a pooled Integrated Public Use Microdata Sample from the American Community Survey for 2008/2009. These reflect the rates for the adults in these income groups, those likely undocumented, and those likely to be legal permanent residents who arrived in the United States in 1996 or later. GROUP 7: ADULTS (AGE 19-64) EMPLOYED IN GOVERNMENT Eligibility and Coverage: This group is not impacted by the Act specifically. However, near universal coverage currently exists for this group and by excluding this group from the remaining categories of adults, we are able to more easily identify local variation in insurance coverage for the adult population. Thus, persons employed in government (for any income category) were first identified and separated from the remaining adults who are classified by household income categories. Currently, 91 percent of this group is covered by some form of health insurance. In the 11

19 limited policy coverage scenario, we assume a 2 percent increase in the percent of the population covered by health insurance (to 93 percent coverage). In the enhanced policy scenario, we assume near universal coverage (98 percent). GROUP 8: ADULTS (AGE 19-64) IN HOUSEHOLDS WITH INCOMES AT OR BELOW 138% OF FPL Pre-Legislation Eligibility and Coverage: Many in this group are not enrolled in health insurance (currently only 48 are covered by some form of health insurance according to estimates from the American Community Survey for 2008/2009). Medicaid eligibility is limited to parents with very low incomes and pregnant women. Thus, if covered by health insurance, individuals in this category are primarily enrolled in employer-sponsored insurance or covered by insurance purchased in the individual market. Post-Legislation Eligibility and Coverage: This group will be eligible for Medicaid. In addition, individuals can receive subsidies and purchase health insurance through the exchange. Assumptions and Methodology for Estimation: In the limited policy scenario, we assume that 71 percent of this group will be covered by some form of health insurance (a 23 percent increase). This is lower than the rate of insurance for Medicaid/CHIP eligible children today. Currently, 76 percent of Medicaid/CHIP eligible children are covered by some form of health insurance (private or public) even though all are eligible for Medicaid or CHIP. Similarly, for various reasons, such as being a young, single, and healthy adult, not all adults eligible for Medicaid will enroll in health insurance. Thus the rate of enrollment in health insurance will likely be lower than that of Medicaid/CHIP eligible children today. The effect of this change in insurance represents a decline of 44 percent of the uninsured. This is lower than the 49 percent decline estimated by Holahan and Headen (2010) for the State of Texas as we assume more limited efforts to enroll individuals in Medicaid. In the enhanced policy scenario, we assume near universal coverage for this population group (98 percent). GROUP 9: ADULTS (AGE 19-64) IN HOUSEHOLDS WITH INCOMES % OF FPL GROUP 10: ADULTS (AGE 19-64) IN HOUSEHOLDS WITH INCOMES BETWEEN % OF FPL Pre-Legislation Eligibility and Coverage: Primarily covered by employer-sponsored insurance or purchased through the individual market (if insured at all). Individuals in this group are not eligible for Medicaid. Post-Legislation Eligibility and Coverage: There are no changes in eligibility for Medicaid (i.e. individuals in these income groups are not eligible for public coverage). Health insurance may be purchased through the exchange and is subsidized for households with incomes at or below 400 percent of FPL. Households in this income range may also be penalized for not having health insurance. 12

20 Assumptions and Methodology for Estimation: Unlike other groups where public health insurance coverage will impact changes to the uninsured and previous expansions in other public programs may provide some guidance on take-up rates for the newly eligible, the rate at which households in this group acquire health insurance is dependent upon decisions made by firms to offer insurance and individuals to purchase insurance for themselves and their families. Currently, according to estimates derived from the 2008/2009 ACS IPUMS, 56 percent of the lower income group is covered by some form of health insurance. In the limited policy coverage scenario, we assume an 11 percent increase in health insurance coverage resulting in a 67 percent coverage rate for this first group. In the enhanced policy scenario, we assume near universal coverage (98 ). In the second group (that is, the group in households with incomes 201 to 400 percent of FPL), we assume a 7 percent increase in the percent insured (to 78 percent of the population group). GROUP 11: ADULTS (AGE 19-64) IN HOUSEHOLDS WITH INCOMES GREATER THAN 400% OF FPL Pre-Legislation Eligibility and Coverage: Adults in households with incomes greater than 400 percent of FPL are not eligible for public health insurance. The primary source of health insurance for this group is through employer-sponsored insurance or through individual policies. Post-Legislation Eligibility and Coverage: No Change to public health insurance rules. In addition, these adults live in households where subsidized exchange based policies will not be available. Assumptions and Methodology for Estimation: This group already has a high rate of health insurance coverage (89 percent). In the limited policy scenario, we assume a slight increase in health insurance coverage as a result of expanded employer coverage and the individual mandate (to 91 percent). We assume near universal coverage in the enhanced policy scenario (98 percent). Adjustments and Evaluation of Estimates Because we are projecting the impacts of policies never before implemented there are no established procedures for assessing the accuracy of these estimates of the impacts of the ACA on Texas or areas within Texas (or other areas of the United States). We have estimated the impacts on the basis of assumptions about how groups of people will be impacted as a result of specific provisions within the legislation. At the state level, we compared our estimates of the overall impacts of the ACA to other estimates of statewide impacts that were prepared by outside groups. Our estimates are consistent with other models of statewide impacts [see Table 3 (Auerbach et al. 2011; Buettgens and Hall 2011; U.S. Congress 2011; Texas Health & Human Services Commission 2010)]. None of the studies cited above have produced estimates of county level impacts of the ACA. In this analysis counties are differentially impacted according to the degree to which these groups are represented within each county. In order to establish reasonable estimates of the impacts of the ACA on these areas, our initial estimates were adjusted in order to be consistent internally and consistent with other county level indicators from other data sources (such as comparisons of government employment from the Bureau of Economic Analysis Regional 13

21 Economic Information System and other data sources). The county estimates were adjusted so that the sum of the impact groups and impact scenarios were consistent with the state-level estimates and consistent within each county. For example, estimates of the elderly insured for all counties combined matched those for the state as a whole. The estimates were also checked for internal consistency by examining the estimated impact on groups according to the geographic characteristics of the county (for instance, we expect higher relative impacts in central counties compared to suburban counties as a result of the low income population). Table 3: Estimates of Health Insurance Coverage for the Nonelderly Population Under ACA Estimation Source Insured Limited Policy Scenario 80 Robert Wood Johnson Moderate Policy Scenario 87 Congressional Budget Office Texas Health & Human Services RAND Enhanced Policy Scenario 94 In addition to the internal consistency checks and adjustments, the data were evaluated by comparing the estimation groups with other county level indicators. These indicators include: county based employment data (government and firms by size), county Medicare enrollment data, ACS county estimates of the population in poverty, ACS county estimates of the foreign-born, and county based current health insurance estimates. Because health insurance coverage is already high for those employed in large firms (100 or more employees) and government, we expect higher current and future health insurance coverage proportional to the county employment in these two groups. In addition, we expect the same high relative health insurance coverage (current and future) proportional to employment in industries with high levels of current health insurance coverage. County level estimates of employment (at the place of work) by firm size and county estimates of government employment are used to check consistency with the impact groups. In addition, metropolitan and micropolitan estimates of firm employment by broad category are compared to the county impact estimates. In addition to the employment indicators, the impact groups are compared to estimates of the foreign-born (to assess whether the undocumented and the legal permanent resident populations are consistent with the estimates of the foreign born); Medicare enrollment data (for consistency with the elderly impact group); county-based estimates of the population in poverty (so that estimates of the groups by poverty status are consistent); and current estimates of the uninsured (to check the overall reasonableness of the overall impacts of the ACA). 14